Chapter One
Introduction
Was ist ĂŒberhaupt ein sprachliches Faktum, das von der Grammatikbe-schreibung zu erfassen ist?
âRenate Steinitz (1981:25)
1. Basic Research Question
1.1. The Linguistic Characterization of Verbal Predicates
This dissertation asks a very fundamental question about the linguistic characterization of verbal predicates, such as laugh, knock at the door, write a letter. Verbal predicates denote situations in a world, and these situations have certain characteristics. In the philosophical literature, predicates are often characterized according to what types of situations they denote (e.g., Ryle 1949, Kenny 1963, Taylor 1985, all exploring ideas originating with Aristotle). And linguists have often looked to the philosophical literature for criteria or classes to use in the linguistic characterization of predicates.
One philosopher whose work has become particularly influential in modern linguistics is Zeno Vendler, who in an essay titled âVerbs and Timesâ divided English verbs (actually, verb phrases) into four temporal-aspectual classes, the well-known states, activities, accomplishments and achievements (Vendler 1957).
- states: know, love (somebody), have, want, be married
- activities: run, push a cart, write, work, walk
- achievements: die, recognize (someone), reach the top, win a race, spot the plane
- accomplishments: run a mile, draw a circle, write a letter, paint a picture, build a house
Vendlerâs criteria for these four classes are loosely semantic, involving how a verb (or VP) can be used, and what kinds of entailment patterns it has. In the rich linguistics literature that followed Vendlerâs paper, these verb classes have been defined and revised in various ways, and formal accounts of them have been developed (e.g., Mourelatos 1978, Dowty 1979, Carlson 1981, Bach 1986, Hinrichs 1985, Moens 1987, Parsons 1990, Piñón 1995). Importantly, much of this work, and most work on tense/aspect, assumes that the Vendler predicate classes, or the characteristics that underlie them, are universal (Comrie 1976, Mourelatos 1978, Dowty 1979, Bach 1981, 1986, Smith 1991, Ehrich 1992, ter Meulen 1995, Musan 2002, etc.). Moreover, a Vendler-type classification is assumed in virtually all work which tries to establish links between temporal-aspectual characteristics of predicates and their grammatical representation. Thus, many theories of argument linking and verb diathesis use lexical semantic or syntactic structures which represent the Vendler predicate classes in some way (e.g., Grimshaw 1990, Jackendoff 1990, 1996, Pustejovsky 1991, Fagan 1992, van Valin 1990, 1993, Tenny 1987, 1994, Borer 1994, 1998, Rapp 1997a,b, Rappaport Hovav and Levin 1998, Ritter and Rosen 1998, 2000, van Hout 2000, Ramchand 2002). To quote Rappaport Hovav and Levin (1998:106):
in most current theories, the aspects of [verb] meaning which are grammatically relevant usually are those which define the various ontological types of events, which correspond roughly to the recognized Vendler-Dowty aspectual classes of verbs.
Thus, almost unnoticed, Vendlerâs philosophical and semantic categories have become grammatical categories, and universal ones at that. What falls by the wayside in this literature is the fundamental question of whether Vendlerâs predicate types really form distinct grammatical classes, in all languages. This often forgotten question is the central concern of my dissertation. I believe it is of crucial importance to ensure that the semantic notions used as the basis of grammatical theories have âgrammatical reality.â This leads to the following general research question, which is the focus of this dissertation:
- (2) General Research Question
- How are temporal-aspectual characteristics of predicates grammatized in natural language?
In the following two sections, I will explain what I mean by âtemporal-aspectual characteristicsâ and by âgrammatized,â and sharpen my research question.
1.2. Temporal-Aspectual Characteristics
âTemporal-aspectual characteristicsâ refer to the characteristics underlying the Vendler verb/predicate classes given in (1), or, in Vendlerâs words, how verbsâ meaning âpresupposes and involves the notion of timeâ (Vendler 1967:97). I now briefly introduce these characteristics, postponing a more careful theoretical discussion until section 4.1.2.
Telicity. The characteristics and associated tests identified by Vendler himself are somewhat unclear (cf. Verkuyl 1989), and subsequent authors, in attempting to streamline Vendlerâs ideas, have typically focused on different subsets of Vendlerâs characteristics. However, there is a broad consensus that one characteristic is telicity; it indicates that a situation has an inherent endpoint or culmination which issues in a result state. The situation itself cannot continue past the inherent endpoint or culmination. For example, the inherent endpoint of the situation denoted by write a letter is the completion of a letter (which results in the existence of said letter). The situation denoted by walk has no such inherent endpointâa walking event can be stopped at an arbitrary point, it is not completed or finished, and there is no result state. Accomplishments and achievements are telic; activities and states are atelic. In English, a convenient telicity test is compatibility with time span adverbials such as in an hour (e.g., â Susi wrote a letter in an hour but # Susi walked in an hour1).
Durativity. There is less agreement on what distinguishes achievements from accomplishments; it might be the presence of a process, it might be durativity, or even agentivity. Vendler himself mentions all three criteria: accomplishments but not achievements are âprocesses,â i.e., they contain âsuccessive phases following one another in timeâ (Vendler 1967:99); achievements but not accomplishments âoccur at a single moment,â i.e., do not have duration (Vendler 1967:103); and, although of secondary importance, accomplishments involve an agent/volition while many achievements do not (Vendler 1967:105f). All three of these are found in other work as well. For example, Dowty (1979) describes the difference between accomplishments and achievements in terms of âtemporally consecutive subsidiary changesâ (p. 181) and/or a causing event (p. 183) present in the former but not the latter. Process is also the notion chosen by Pustejovsky (1991) to distinguish accomplishments from achievements, and states from activities. Smith (1991), on the other hand, proposes that the relevant characteristic is durativity, i.e., whether a situation takes time (has duration) or occurs in an instant. For example, write a letter (accomplishment) usually denotes a situation that takes some time, whereas recognize someone (achievement) denotes a situation that happens in an instant. In English, durative predicates are compatible with a duration adverbial such as for an hour or with the verb stop: â Susi stopped writing a letter but # Susi stopped recognizing him.
Following Smith (1991), I choose durativity rather than process or agentivity as the relevant characteristic. I make this choice because durativity is a truly temporal notion which can be clearly defined and easily tested. Agentivity arguably is not a temporal-aspectual notion but a thematic one, having to do with the presence of a volitional, sentient, possibly causing, event participant rather than with the temporal contour of an event. As far as I am aware, authors which mention agentivity do point out that it is not really aspectual (e.g., Vendler 1957, Pustejovsky 1991), and I will disregard agentivity in this study.
I also reject the characteristic of process, for the following reasons. First, process is a mixed notion which has to do with agentivity and causation, thus overlapping with the thematic domain. This is best seen by looking at attempts to define process. For example, Dowtyâs (1979:109 and 118) definitions in terms of CAUSE and DO invoke causation and agentivity, respectively, and for Pustejovsky the presence of a process in accomplishments correlates with agentivity rather than with something inherently aspectual (1991:59). Second, it is difficult to come by tests for process which do not interact with agentivity, particularly in German (see Wilhelm 2000), one of the languages studied here. Thus, using the notion of process may not give very clear results. Finally, the characteristic of process, or successive phases, is not always easy to perceive. For example, what is the process component in the accomplishments wilt or decide? Similarly, many activities do not have a perceptible process, compare sleep (Dowty 1979), enjoy (Smith 1999), fall, float. Because of the difficulty of identifying successive stages, Smith (1999) proposes to formalize this component (which she calls dynamism, but which in my view is indistinguishable from process) in terms of its mapping onto time. In other words, Smith appeals to a temporal definition, a definition in terms of durativity.
In conclusion, process apparently cannot be described, defined, or tested without reference to agentivity/causation or durativity. I therefore choose the simpler and more clearly temporal-aspectual notion of durativity in my study on the grammatization of temporal-aspectual characteristics of predicates. As we shall see in the course of this dissertation, the study of durativity gives clean and interesting results.2
Stativity. Finally, although Vendler puts states in a class with achievements, because purportedly they both lack a process and are true of a single moment, it has since become clear that states are a class all by themselves. As opposed to nonstative situations, also called events, stative situations have no dynamicsâthey continue without input of energy, and have no internal temporal structureâthey consist of undifferentiated moments. This difference becomes clear when comparing stative know (French) and nonstative walk or write a letter. There is a vast literature on states, types of states, and their semantic and syntactic representation (Carlson 1977, Dowty 1979, Kratzer 1995, Parsons 1990, Smith 1991, to name just a few).
The temporal-aspectual predicate classes and underlying characteristics I thus arrive at are based on Smith (1991) and summarized in (3). Smith calls the predicate classes âsituation types,â and I will follow this terminology, but sometimes call nonstative situations âeventsâ (Mourelatos 1978, Parsons 1990). Notice that Smith departs from Vendler in making a durativity distinction not only among telic, but also among atelic situations. Activities are atelic and durative, semelfactives are atelic and nondurative. Such a differentiation among atelics is only consistent, and I assume the existence of the situation type âsemelfactiveâ in this dissertation.
- (3) Temporal-aspectual classification of predicates, based on Smith (1991)
SITUATION TYPE | STATIVE | DURATIVE | TELIC | ENGLISH EXAMPLES |
State | + | + | n/a | know the answer, own |
Activity | â | + | â | run, read |
Accomplishment | â | + | + | build a house, read a book |
Semelfactive | â | â | â | tap, knock |
Achievement | â | â | + | find the key, reach the top |
To keep this dissertation manageable, I will put aside states and stativity, and focus on durativity and telicity, which distinguish the different types of nonstative situations from each other. I thus sharpen my research question as follows:
- (4) Revised Research Question
- How are telicity and durativity grammatized in natural language?
1.3. Grammatization
By âgrammatizedâ I mean that the explanation of some grammatical phenomenon or contrast must require reference to these characteristics, i.e., to telicity or durativity. Thus, I am not primarily interested in how temporal-aspectual characteristics of predicates are represented in lexical or sentential meaning in general, but in which characteristics form the basis of a morphological or a syntactic phenomenon or contrast. In other words, I mean by âgrammaticalâ something that can be seen in the forms of words or phrases of a given language. I will assume the following definition of âgrammatization:â
- (5) Grammatization
- A semantic notion/feature evident in a productive morphosyntactic contrast is grammatized.
Here âmorphosyntaxâ is used in a wide sense, referring to either a morphological or a syntactic contrast.3 The contrast has to be productive, however, i.e., it cannot just be an isolated, irregular or frozen phenomenon. For example, it is sometimes suggested that German has a pair of verbal suffixes -l/-r which impart durative, or rather iterative, meaning on a verb, as in klappern ârattle, clatterâ < klappen âfold rigid object (lid, door)â (Abraham, p.c., Duden 1984, FlĂ€mig 1965). However, it is far from clear what the meaning of these suffixes is, or whether they are productive synchronically. As to meaning, -l/-r have a range of semantic effects, of which durative/itera...